Broadcom's custom AI chip business is growing faster than its own guidance, with semiconductor revenue hitting $10.8 billion in the fiscal second quarter — a 143% year-over-year surge that positions the company as the primary alternative to Nvidia in the hyperscaler supply chain.
Broadcom's custom AI chip business is growing faster than its own guidance, with semiconductor revenue hitting $10.8 billion in the fiscal second quarter — a 143% year-over-year surge that positions the company as the primary alternative to Nvidia in the hyperscaler supply chain.
The shift toward custom silicon is no longer theoretical. Broadcom's AI semiconductor revenue rose 143% year over year to $10.8 billion in fiscal 2026's second quarter, driven by demand for custom accelerators and networking hardware from the world's largest data center operators. Consolidated revenue climbed 48% to a record $22.2 billion, while adjusted EBITDA rose 52% to $15.2 billion, producing a 69% margin.
"Hyperscalers are increasingly turning to custom AI silicon as they look for alternatives to Nvidia's GPUs, and Broadcom is the primary beneficiary of that shift," the company said in its earnings release. Broadcom has partnered with Google, Anthropic and OpenAI to develop custom chips, giving it a pipeline that extends well beyond a single customer.
The company generated $10.3 billion in free cash flow during the quarter despite just $231 million in capital expenditure, reflecting its capital-light, outsourced manufacturing model. Infrastructure software, anchored by the VMware acquisition, contributed $7.2 billion in revenue — 32% of total sales — though growth in that segment has slowed to 9% year over year after a 26% surge in fiscal 2025 driven by post-acquisition price increases.
Why Custom Chips Are Reshaping the AI Supply Chain
Broadcom's growth reflects a structural shift in how hyperscalers procure AI compute. Rather than relying exclusively on Nvidia's general-purpose GPUs, companies like Google, Amazon and Meta are designing custom accelerators — known as ASICs — tailored to their specific inference workloads. Broadcom's role is to design and integrate these chips, then manage the networking fabric that connects them.
The economics are compelling. Custom chips can reduce total cost of ownership by 30% to 50% for inference workloads compared with off-the-shelf GPUs, according to industry estimates. That cost advantage is driving adoption even as Nvidia's data center revenue reached $75.2 billion in its most recent quarter — roughly seven times Broadcom's AI chip sales.
Broadcom's AI semiconductor revenue already exceeds the combined data center sales of AMD ($5.8 billion) and Intel ($5.1 billion) in their most recent quarters. Management expects that gap to widen, guiding for Q3 AI semiconductor revenue of $16 billion — growth of more than 200% year over year — and consolidated revenue of $29.4 billion, up 84%.
Valuation and the Investor Calculus
Broadcom shares trade at roughly 68 times earnings, more than double Nvidia's 32x multiple. That premium reflects investor expectations that Broadcom's earnings have not yet peaked — that its custom chip business can continue expanding as more hyperscalers diversify away from single-source suppliers.
A consensus of 42 analysts rates Broadcom a strong buy, with the highest price target implying 56% upside from current levels. The stock has rallied 63% over the past 52 weeks, trading in a range of $244 to $495, though it has slipped about 2% in the past month after management's Q3 guidance failed to clear elevated investor expectations.
For investors, the question is whether Broadcom can sustain this trajectory. The company's Q3 guidance implies AI semiconductor revenue will account for more than half of total sales for the first time. If hyperscalers continue favoring custom silicon — and Broadcom's partnerships with Google, Anthropic and OpenAI suggest they will — the company could emerge as the second-largest AI chip supplier by revenue, trailing only Nvidia. The June quarter results, due in late August, will provide the next data point on whether that trajectory remains intact.
This article is for informational purposes only and does not constitute investment advice.
